BACKGROUND IMAGE: iSTOCK/GETTY IMAGES
SAN FRANCISCO -- It's been two years since Salesforce formally introduced the Einstein AI platform that is supposed to help users find hidden insights within their data. At Dreamforce 18, some customers described how they have integrated Einstein AI capabilities into their business processes.
Every Salesforce product has received some kind of Einstein AI upgrade, with more features rolled out with each product release. Companies like tire provider Michelin and Dublin-based recruiting company CPL are among the Salesforce customers that have found ways to use Einstein AI to help with productivity and revenue growth.
"AI is there to enhance the human input," said Danielle DeLozier, global product owner for Service Cloud at Michelin, based in Clermont-Ferrand, France. "We realized a few years ago that customer expectations have increased drastically, and we wanted that 360-degree view of our customers across all areas that we work in."
Michelin implemented some Salesforce AI functionality in January to help address service issues in the field with customers and Michelin's main product -- tires. DeLozier implemented Einstein Vision, a photo capture feature that uses AI to identify details of an image and diagnose the problem.
"We were taking pictures of tires in the field to see if we could identify issues or usage trends to help resolve those issues," DeLozier said. "We saw what the value Einstein and predictive analytics could bring to our customers and employees."
But Einstein AI isn't a simple plug and play; there are several aspects of the business that need to be in tune to make sure an artificial intelligence implementation works. Organizations need to make sure their data sets are clean and accessible -- and even then, it can take some convincing to get stakeholders or management on board with a new way of doing things, especially when that new way comes with a big price tag.
"AI is not as proven as some of the [other] features Salesforce has," DeLozier said. "When working with the business, it has been a challenge to prioritize this initiative over things we know work; we know live message works. We had to work with stakeholders and say, 'It's worth taking the risk and it might fail, but we can learn from that.'"
While Michelin's DeLozier was focused on improving customer service with Salesforce AI, CPL CIO Kevin Sweeney has implemented AI capabilities to help his company's partners find the right employee for the right position. As an international recruiting company, CPL has more than 1.3 million candidates in its database and receives roughly 40,000 resumes per month.
Prior to implementing Einstein AI shortly after it was unveiled at Dreamforce 2016, Sweeney said the candidate-employer connection was done primarily through Excel and search keys.
"This was not fun in Excel; it was doable, but challenging," Sweeney said, adding that artificial intelligence has its own challenges, as well. "In AI, you effectively don't know what the right answer is, and it takes some time to go through that model."
CPL has used Salesforce for more than eight years, and that breadth of data is useful when implementing Einstein AI.
"We wanted to know if we could predict a candidate's fit based on our history," Sweeney said. "We know, for example, that this candidate worked for a specific company or job title, so is there a way of identifying those candidates that are a good fit, but haven't been contacted."
Einstein AI was added as an aid in addition to the processes CPL already had to find candidates, he said.
'An inherent complexity'
Both Michelin and CPL are ahead of the curve with AI implementation. According to a survey of over 3,000 business executives published in September 2018 by MIT Sloan and the Boston Consulting Group, AI is in use within a small percentage of companies -- just under one-fifth of those polled are pioneering AI.
"AI is fundamentally different from other IT systems," said Philip Cooper, vice president of product go-to-market for Einstein at Salesforce. "AI is nondeterministic -- we're letting it learn from data and not be rules-based."
That idea clashes with how humans think, Cooper said, adding that humans are rules- and fact-based and that leaving something up to probability is difficult.
"If you change one data point, it can change the rest of the model," Cooper said. "It's an inherent complexity."